Heroku MCP Server: Your AI Assistant Gets Platform Powers

Stop switching between your AI coding assistant and the Heroku CLI. This official MCP server gives Claude, Cursor, and other AI tools direct access to your entire Heroku infrastructure through natural language.

Why This Changes Your Workflow

You're already using AI assistants for code. Now they can manage your deployments, scale your dynos, query your databases, and troubleshoot issues without you leaving the conversation. No more copy-pasting commands or switching contexts.

Before: "Let me check the dyno status... opens terminal heroku ps -a myapp... copies output back to AI"

After: "Check if my production app is having memory issues and scale it if needed."

The AI handles the entire workflow—checking dyno status, analyzing metrics, and scaling appropriately.

What You Get

This isn't just another API wrapper. It's a complete Heroku platform interface that understands context:

Application Lifecycle Management

• Create, deploy, and manage apps with full configuration control
• Handle team deployments and private spaces
• Execute one-off dynos for testing, scripts, or temporary workloads
• Manage maintenance windows and app transfers

Production Operations

• Scale dynos up/down or resize them based on performance needs
• Restart specific processes or entire formations
• Stream and analyze application logs
• Monitor and manage PostgreSQL databases with advanced tooling

Advanced Database Operations

The PostgreSQL integration goes beyond basic queries:

• Execute complex SQL with full transaction support
• Identify performance bottlenecks with pg_outliers
• Debug blocking queries with pg_locks
• Manage credentials and handle database upgrades
• Schedule and restore backups

Pipeline and Team Management

• Promote releases through your deployment pipeline
• Manage team access and private spaces
• Coordinate add-on provisioning across environments

Real-World Scenarios

Incident Response: "My app is responding slowly. Check the database for long-running queries and high resource usage, then show me the recent logs."

Deployment Management: "Promote the staging build to production, then scale the web dynos to 3 and worker dynos to 2."

Database Investigation: "Find the slowest queries from the last hour and check if there are any blocking transactions."

Development Setup: "Create a new app for this feature branch, add PostgreSQL and Redis, then deploy the current code."

Setup Takes 30 Seconds

Two options, both dead simple:

Option 1: Use Your Existing Heroku CLI Session

{
  "mcpServers": {
    "heroku": {
      "command": "heroku mcp:start"
    }
  }
}

Uses your current Heroku authentication—no API key management required.

Option 2: Direct NPX Installation

{
  "mcpServers": {
    "heroku": {
      "command": "npx -y @heroku/mcp-server",
      "env": {
        "HEROKU_API_KEY": "your-token-here"
      }
    }
  }
}

Works with Claude Desktop, Cursor, Zed, Windsurf, VSCode, and other MCP-compatible clients.

Integration Examples

Debugging Production Issues:

"My users are reporting 500 errors. Check the app logs for the last 10 minutes, 
look at the current dyno status, and see if there are any database connection issues."

Deployment Workflow:

"Deploy the current branch to a new staging app, run the database migrations, 
then promote to production if the health checks pass."

Performance Optimization:

"Analyze the database performance, show me the top 5 resource-intensive queries, 
and recommend scaling adjustments based on current dyno metrics."

Built by Heroku, For Developers

This is official Heroku tooling, not a community project. It's actively maintained with the same reliability standards as the Heroku CLI itself. The MCP implementation follows best practices and includes comprehensive error handling.

The server supports all Heroku plan types—Common Runtime, Cedar Private Spaces, and Shield Spaces—so it works regardless of your infrastructure setup.

Ready to Start?

Add this to your MCP client configuration and start managing your Heroku infrastructure through natural language. Your AI assistant just became your DevOps partner.

Quick Install with Claude Desktop or check the full configuration guide for other clients.